<ul><li>Anthropic is focusing on agent safety and their approach to tackle it in the coming years.</li><li>They emphasize the importance of measurement, model performance, and compute poured into model training over time.</li><li>They introduce RSP (Responsible Scaling Policy) modeled on bio safety levels to precommit to safety rules and be publicly accountable.</li><li>They highlight the need for defensive layers and mention their Long term benefit trust policy, prioritizing safety and public welfare.</li></ul>

Here’s how Anthropic is approaching Agent safety…..

Discover more